Wednesday, 20 June 2012

Middling standards, being mean, and how every child matters

Right, let's rattle some cages. That headline probably makes you think, as we pass through exam season and await the grades that August brings, that this will be another blog / movement / initiative / mentoring scheme which is about pushing up "satisfactory" or "bog-standard" schools / teachers / students through forceful management and monitoring, so that your class / department / school / federation / pan-galactic empire achieves 100% A*-C. You will be disappointed with what follows as I intend to tell you I pay little attention to the C/D borderline and sometimes let students fail even though I know I could get them C. But hear me out - and first I need to go a little around the houses in some theory.

School value-added analyses continue to use the mean as their average when in fact they should use the median. How a mathematical distinction which should be obvious to a level 4 student slips by 400,000 teachers every year quite escapes me. Let me explain. Ten of 30 students in your class hit their data targets. Ten beat it - usually by one grade. Ten miss it - seven of them by one grade and three recalcitrant sorts refuse to do anything work-wise and each miss by three to six grades. Use the median and your value-added is zero - you're exactly performing to target (if indeed student grades can be used to measure teachers, but let's presume so for today.) Use the mean and the data is skewed heavily by those three refusers: you're at minus a third of a grade on average or so and hovering on statistical underperformance. Certainly you are going to have a difficult performance management meeting this year. But the same number of children beat target as fell short. But two of the three problem students are problems right across the school - and one barely attends. A mean average, as any year 8 maths student should know, is skewed by a small number of extreme data points. And whilst it is almost impossible for an individual student to get more than +2 grades above data, it is easy - if you resist, refuse to study, or absent yourself - to get far worse.

The appeal goes that "every child matters" - you can't let those students fail, even if they represent only 10% of your class, even if they are choosing to fight you. Somehow you have to get through - every child matters, and their future is in your hands even if their past is a problem. Fairness, goes the argument, is a society in which we give up on no child. This is idealistically true but practically limited, and here's why: ask yourself how much more time you spend on those students than on the diligent (and equally deserving in a different way) who embrace every task you set and who want help moving from B to A - or A*. You're not allowed to answer that you help them too - because you don't, if you're endlessly chasing the problems to drag them up to C. You're lying to yourself. And you might need to ignore the instruction to get everyone Cs and back a more natural justice approach - perhaps a somewhat Tory approach of telling people to sort themselves out before they're entitled to more help. Teachers' time is a finite resource and a decision to devote it to one cause is, ipso facto, a decision to reject another: to "save" one child is by that very choice to abandon another - and just because the abandoned don't complain as loudly doesn't make it any better a choice.

The C/D borderline is not the be-all and end-all and principled teachers, confident of their purpose, should ignore it. Focus on excellence and you will get results - and I mean this at all levels of the academic scale, and in all contexts. Focus on results and all you will get is narrow-purpose minimal-competence drilling - mediocrity by any other name. I had a conversation almost a decade back with a different Head of Department colleague who got 100% A*-C in their department - more than three-quarters of them C. I got about 90% as 3 students in my department missed C, and she held forth that she had "beaten" me. But my department's median grade was A and hers was C. My value-added median was +2 over data targets and hers was 0. In addition she had managed to massage a cohort into universal middling values which leave employers and universities completely unable to distinguish talent and commitment from spoon-fed apathy. She and her staff had devoted 80% of her time to 20% of her students and settled for moderate competence from the highly able. My department had 6 A*s and strong A-level take-up; hers had one and little, respectively. I'll take my 90% A*-C rate over her 100%, thank you. The point is this: by all means invest that reasonable amount of extra time, sympathy, support, and pushing in those struggling to find a way to love the subject and engage with it; but in the end, stop depriving others who deserve your attention too, at a different level, and distribute yourself more evenly to all. If fairness is a society in which we give up on no child, why are you ignoring the potential A* students on the other side of the room? (And the F/G who could reach E too.) Why are you investing more time in catch-up sessions than extension opportunities? If EVERY child matters - that means the able and well-behaved too. Sometimes those who continue to resist, refuse and who want to fail need to be allowed to do so. It's not giving up. It's giving your best, and more time than they're due, and then leaving them the consequences of their choices. Keep the lines of communication open (often the recognition that, although you're there and waiting to work with them, you will no longer continue to be at their beck and call endlessly, or that you have stopped prioritising them, is the very stimulus that wakes them up and kick-starts a late recovery) and hope that, even if they don't turn this round in time, they'll learn the lesson in time for a resit or the next stage of qualifications. Remember we are teaching children for life, about self-application. Every child matters: go spend some time with that potential A* over there instead. Don't be happy about his high grade B essay comparing Brutus and Antony's speeches in Act 3 of Julius Caesar; ask him why he isn't both cross-comparing them and also comparing them to the previous evidence of the two protagonists' private character and public image - and perhaps also to other great speeches of Shakespeare. Letting 2-3 students each year fail if their engagement level proves they intend to, and being conspicuous in not bending over backwards to do it all for them, can have powerful effects in motivating not just those of high potential but the middling too: if they know that both (a) people will be allowed to get their just deserts - high or low and that (b) you'll invest time in anyone and it's their responsibility to make what they can of that - suddenly a previously ignored majority can fly. Try to get through to the demotivated, and always provide every help to the weak, but give everyone the time they deserve; and make sure parents get the message if their child is under-applying - and that it is THEIR business (students AND parents), and not yours, to fix that.

And this is why I refuse to use the mean in data analyses. If twenty students in a class of thirty-one beat their target grade (as happened in my GCSE class last year), most by one grade, I'm not about to let two wilfully apathetic refusers on target-minus-four cancel that out. So I'm taking the median instead, which treats each child underperforming as the same as each child overperforming - what matters is how many under/over perform, not by how far. Because every child matters - so a mean value-added analysis which scores those two refusers as four times as important as each of the twenty who worked damned hard to overperform - is just downright offensive. Means and medians are averages - are ways of working out "the middle" - well, what I'm saying is that if you use the mean, your focus is on middling values - in the derogatory sense - on block-scoring mediocre, minimal-competence C grades. It's offensively preferential, it panders to resisters over the deserving, it sends the wrong message about quality in education, and it is useless to employers. I reject the mean, and the obsession with C/D, on principle - and don't be surprised that I therefore almost always beat, in practice, those who champion it solely.

At this point, depressingly, there are probably still some alarmed Senior Leaders out there reading this and thinking "He's dangerous, I'd fire him", or at least "I'd get him to see why he's wrong" (which means, "I'd make him change back to focusing on the C/D borderline and if not, then I'd fire him.") And it is these people, and not my approach, that is the problem. In terms of getting students A* grades last year, I was seventy-five times as effective as the average core subject teacher at my last school. SEVENTY-FIVE TIMES. And I am not walking on water; I am not especially brilliant; I simply reject a dogma of triage about getting everyone C both on principle and because in practice it always results in dumbing-down. And there are people who think the problem is my approach? Senior Leaders - I ask you: who wouldn't want a teacher who gets these results? (And for the record, I have never, ever, taught in a selective context, and the school I am referring to last year was only 3 years ago in Special Measures and on Ed Balls' famous hit list. The course was sat in 60% of the recommended course time. I invite you to challenge my credentials on social justice vs elitism and on quality vs context.)

So ask yourself whether you're dividing your time fairly. And next time you're asked for a value-added analysis, use the median and not the mean, and strongly challenge anyone who uses a mean instead. It's OK that some students fail. It's not OK that many students don't reach their fullest potential because we're obsessing about the far fewer who might fail. Because every child matters - not some more than others. Think what it really means.